Re: [Neo] Graph Algo Problems
Hi Peter, I have the centrality stuff working now, no further help required. :) My nodeset had to be larger. -Todd On Mar 1, 2010, at 10:02 PM, Todd Stavish t...@stavi.sh wrote: Hi Everyone, I am running into some null pointer exceptions when I try to feed things into some of the graph algos. The test cases can be found in the link below. It is all maven based and the pre-built graph is included. Queries19 (Floyd-Warshall), 20 (Eigenvector Centrality with the Arnoldi iteration), and 21 (Eigenvector Centrality with the Power method) are the problematic ones. To run each query, change the query number in the configuration.properties in the resources directory. Then, mvn compile, mvn exec:exec. http://www.stavi.sh/backend.zip I constrained the sub-graph to be the smallest possible one. I also tried setting the max iterations and calling calculate before the getCentrality call. I followed the same pattern I used for the BFS-Centrality and Dijikstra for shortest path (which seem to be working). Not sure what I am doing wrong here. Any help would be appreciated. Thanks, Todd ___ Neo mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user
Re: [Neo] Graph Algo Problems
Ok, great, I am just now looking at it and noticed a NPE in Node topicNodes = referenceNode.getSingleRelationship(RelationshipTypes.TOPICS, Direction.OUTGOING).getEndNode(); The referenceNode.getSingleRelationship(RelationshipTypes.TOPICS, Direction.OUTGOING) is returning NULL ... Cheers, /peter neubauer COO and Sales, Neo Technology GTalk: neubauer.peter Skype peter.neubauer Phone +46 704 106975 LinkedIn http://www.linkedin.com/in/neubauer Twitter http://twitter.com/peterneubauer http://www.neo4j.org - Your high performance graph database. http://www.tinkerpop.com - Processing for Internet-scale graphs. http://www.thoughtmade.com - Scandinavias coolest Bring-a-Thing party. On Wed, Mar 3, 2010 at 6:55 PM, Todd Stavish toddstav...@gmail.com wrote: Hi Peter, I have the centrality stuff working now, no further help required. :) My nodeset had to be larger. -Todd On Mar 1, 2010, at 10:02 PM, Todd Stavish t...@stavi.sh wrote: Hi Everyone, I am running into some null pointer exceptions when I try to feed things into some of the graph algos. The test cases can be found in the link below. It is all maven based and the pre-built graph is included. Queries19 (Floyd-Warshall), 20 (Eigenvector Centrality with the Arnoldi iteration), and 21 (Eigenvector Centrality with the Power method) are the problematic ones. To run each query, change the query number in the configuration.properties in the resources directory. Then, mvn compile, mvn exec:exec. http://www.stavi.sh/backend.zip I constrained the sub-graph to be the smallest possible one. I also tried setting the max iterations and calling calculate before the getCentrality call. I followed the same pattern I used for the BFS-Centrality and Dijikstra for shortest path (which seem to be working). Not sure what I am doing wrong here. Any help would be appreciated. Thanks, Todd ___ Neo mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user ___ Neo mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user
Re: [Neo] Graph Algo Problems
Those exceptions seem like it might not be setup properly. No matter, I changed the implementation. I am seeing different centrality calculations for Arnoldi versus Power. I tweaked the iterations to see if it made a difference. Is it the precision setting causing the difference maybe? The target and nodeset are the same for each. -Todd On Mar 3, 2010, at 11:04 AM, Peter Neubauer peter.neuba...@jayway.com wrote: Ok, great, I am just now looking at it and noticed a NPE in Node topicNodes = referenceNode.getSingleRelationship(RelationshipTypes.TOPICS, Direction.OUTGOING).getEndNode(); The referenceNode.getSingleRelationship(RelationshipTypes.TOPICS, Direction.OUTGOING) is returning NULL ... Cheers, /peter neubauer COO and Sales, Neo Technology GTalk: neubauer.peter Skype peter.neubauer Phone +46 704 106975 LinkedIn http://www.linkedin.com/in/neubauer Twitter http://twitter.com/peterneubauer http://www.neo4j.org - Your high performance graph database. http://www.tinkerpop.com - Processing for Internet-scale graphs. http://www.thoughtmade.com - Scandinavias coolest Bring-a-Thing party. On Wed, Mar 3, 2010 at 6:55 PM, Todd Stavish toddstav...@gmail.com wrote: Hi Peter, I have the centrality stuff working now, no further help required. :) My nodeset had to be larger. -Todd On Mar 1, 2010, at 10:02 PM, Todd Stavish t...@stavi.sh wrote: Hi Everyone, I am running into some null pointer exceptions when I try to feed things into some of the graph algos. The test cases can be found in the link below. It is all maven based and the pre-built graph is included. Queries19 (Floyd-Warshall), 20 (Eigenvector Centrality with the Arnoldi iteration), and 21 (Eigenvector Centrality with the Power method) are the problematic ones. To run each query, change the query number in the configuration.properties in the resources directory. Then, mvn compile, mvn exec:exec. http://www.stavi.sh/backend.zip I constrained the sub-graph to be the smallest possible one. I also tried setting the max iterations and calling calculate before the getCentrality call. I followed the same pattern I used for the BFS-Centrality and Dijikstra for shortest path (which seem to be working). Not sure what I am doing wrong here. Any help would be appreciated. Thanks, Todd ___ Neo mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user ___ Neo mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user ___ Neo mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user
Re: [Neo] Graph Algo Problems
Thanks Todd for the pointers! Is this absolutely critical for you? Otherwise, we will put it into our backlog and do as time permits. Great you got testcases, so I'm sure we should be able to sort it out. The graph algos are written some time ago and would need some general review - we are not quite satisfied with the semantics and logic of things. Meanwhile, another possibility would be to look at the JUNG algos via Gremlin, see http://wiki.github.com/tinkerpop/gremlin/working-with-jung-algorithms. They do not use traversers, but the algos themselfes should be very good and otherwise performance is prolly ok. Cheers, /peter neubauer COO and Sales, Neo Technology GTalk: neubauer.peter Skype peter.neubauer Phone +46 704 106975 LinkedIn http://www.linkedin.com/in/neubauer Twitter http://twitter.com/peterneubauer http://www.neo4j.org - Your high performance graph database. http://www.tinkerpop.com - Processing for Internet-scale graphs. http://www.thoughtmade.com - Scandinavias coolest Bring-a-Thing party. On Tue, Mar 2, 2010 at 6:02 AM, Todd Stavish t...@stavi.sh wrote: Hi Everyone, I am running into some null pointer exceptions when I try to feed things into some of the graph algos. The test cases can be found in the link below. It is all maven based and the pre-built graph is included. Queries19 (Floyd-Warshall), 20 (Eigenvector Centrality with the Arnoldi iteration), and 21 (Eigenvector Centrality with the Power method) are the problematic ones. To run each query, change the query number in the configuration.properties in the resources directory. Then, mvn compile, mvn exec:exec. http://www.stavi.sh/backend.zip I constrained the sub-graph to be the smallest possible one. I also tried setting the max iterations and calling calculate before the getCentrality call. I followed the same pattern I used for the BFS-Centrality and Dijikstra for shortest path (which seem to be working). Not sure what I am doing wrong here. Any help would be appreciated. Thanks, Todd ___ Neo mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user ___ Neo mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user
Re: [Neo] Graph Algo Problems
Hi Peter, Thanks for the help. It is critical. I need to provide feedback to my customer on the suitability of using the in-built Neo4J graph algos. Or if someone has a working testcase of Floyd-Warshall or either Eigenvector implementation, I could use that to compare with my code / graph. Thanks, Todd On Tue, Mar 2, 2010 at 3:10 AM, Peter Neubauer neubauer.pe...@gmail.com wrote: Thanks Todd for the pointers! Is this absolutely critical for you? Otherwise, we will put it into our backlog and do as time permits. Great you got testcases, so I'm sure we should be able to sort it out. The graph algos are written some time ago and would need some general review - we are not quite satisfied with the semantics and logic of things. Meanwhile, another possibility would be to look at the JUNG algos via Gremlin, see http://wiki.github.com/tinkerpop/gremlin/working-with-jung-algorithms. They do not use traversers, but the algos themselfes should be very good and otherwise performance is prolly ok. Cheers, /peter neubauer COO and Sales, Neo Technology GTalk: neubauer.peter Skype peter.neubauer Phone +46 704 106975 LinkedIn http://www.linkedin.com/in/neubauer Twitter http://twitter.com/peterneubauer http://www.neo4j.org - Your high performance graph database. http://www.tinkerpop.com - Processing for Internet-scale graphs. http://www.thoughtmade.com - Scandinavias coolest Bring-a-Thing party. On Tue, Mar 2, 2010 at 6:02 AM, Todd Stavish t...@stavi.sh wrote: Hi Everyone, I am running into some null pointer exceptions when I try to feed things into some of the graph algos. The test cases can be found in the link below. It is all maven based and the pre-built graph is included. Queries19 (Floyd-Warshall), 20 (Eigenvector Centrality with the Arnoldi iteration), and 21 (Eigenvector Centrality with the Power method) are the problematic ones. To run each query, change the query number in the configuration.properties in the resources directory. Then, mvn compile, mvn exec:exec. http://www.stavi.sh/backend.zip I constrained the sub-graph to be the smallest possible one. I also tried setting the max iterations and calling calculate before the getCentrality call. I followed the same pattern I used for the BFS-Centrality and Dijikstra for shortest path (which seem to be working). Not sure what I am doing wrong here. Any help would be appreciated. Thanks, Todd ___ Neo mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user ___ Neo mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user ___ Neo mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user
